Import libraries

0.1. Make a folder to save graph

1. Load data

1.1. Load processed oracle object

Load the oracle object (in-house E12.5 scRNA-seq) that was created previously.

1.2. Load inferred GRNs

Load previously generated GRNs based on integrated snATAC-seq and scRNA-seq of E12.5 mouse cerebellum.

2. Make predictive models for simulation

Fit ridge regression models.

3. in silico Perturbation-simulation

Here, we simulate gene knockout of Hes1, Hes5, Ptf1a, or Atoh1 on cell fate specification, by setting it expression level to 0.0.

Simulate double KO of Hes1 and Hes5

Here we simulate double KO of Hes1 and Hes5.

Make another Saneky diagram with different cluster units.

GOF analysis

Here, we simulate gain-of-function experiment by increasing 2 fold of the endogenous Hes1, Hes5, Ptf1a, or Atoh1.

Simulate ectopic expression of Atoh1 from the Ptf1a locus, or vice versa

Here, we simulate two knock-in experiments, in which Atoh1 is knockd into the Ptf1a locus, or vice versa. To this end, we calculate the sum of the expression of Ptf1a and Atoh1, and added to either Ptf1a (ectopic Ptf1a) or Atoh1 (ectopic Atoh1).

In the following output, we can see that Atoh1 is now ectopically expressed in Ptf1-expressing cells as expected as in the Atoh1-Ptf1a KI cerebellum.

To fully simulate Atoh1-Ptf1a KI, which will abolish Ptf1a expression, we set Ptf1a expression level to 0.

Simulate Ptf1a KI into the Atoh1 locus

As shown in the following output, Ptf1a is now ectopically expressed in the Atoh1 domain.